CN111033289A

CN111033289A - Diagnostic device for coil and diagnostic method for coil

Info

Publication number: CN111033289A
Application number: CN201880051966.4A
Authority: CN
Inventors: 木村英明
Original assignee: Aisin AW Co Ltd
Current assignee: Aisin Corp
Priority date: 2017-08-31
Filing date: 2018-08-31
Publication date: 2020-04-17
Anticipated expiration: 2038-08-31
Also published as: US11156667B2; US20200158790A1; CN111033289B; JP6849080B2; JPWO2019045044A1; WO2019045044A1

Abstract

The present invention relates to a diagnostic device and a diagnostic method for a coil. The quality of the coil is diagnosed in a wider target range of the coil based on a response voltage obtained by applying the surge voltage. A diagnostic device (1) for a coil (30) is provided with: a voltage applying unit (2) for applying a surge voltage to the coil (30); a response voltage detection unit (3) that detects a response voltage from the coil (30) to the surge voltage; an index calculation unit (5) that calculates a determination index indicating the electrical characteristic of the coil (30) on the basis of the response voltage; and a determination unit (6) for determining whether or not the target coil is abnormal by comparing a determination index of a reference coil, which is a normal coil, with a determination index of the target coil, which is a diagnostic coil (30), wherein at least one of a zero cross point at which the response voltage crosses the reference voltage and peak voltages on the positive side and the negative side of the response voltage is used as the determination index in addition to the circuit constant of the coil (30).

Description

Diagnostic device for coil and diagnostic method for coil

Technical Field

The present invention relates to a diagnostic device for diagnosing whether or not an abnormal coil is present.

Background

A technique is known in which a surge voltage is applied to a coil of a rotating electric machine, a transformer, or the like, and the response voltage is observed to diagnose whether the coil is good or bad. For example, japanese patent laid-open No. 2012-242377 determines characteristic quantities (LC and RC) of the coil based on the response voltage, a differential voltage obtained by differentiating the response voltage, and a second order differential voltage (second order derivative) obtained by further differentiating the differential voltage, thereby diagnosing whether the coil is good or bad. Here, LC is a product of inductance and capacitance, and RC is a product of resistance (resistance component) and capacitance ([0014] to [0015], and the like).

When an abnormality occurs in a coil to be diagnosed, the insulation between conductors constituting the coil is reduced. Even if the insulation between conductors at the same potential in the coil is reduced, the electrical performance is hardly affected, and it is difficult to diagnose whether the coil is good or bad based on the characteristic quantities (LC and RC) related to the circuit constant. Such a decrease in insulation property hardly affects electrical performance, and thus has few practical problems. However, from the viewpoint of production control, there is a cause of such insulation failure occurring in the coil, and it is preferable to pay attention also to the possibility of insulation failure occurring at a position that affects electrical performance. Therefore, it is desirable to appropriately detect such a poor insulation at a position having the same potential.

Patent document 1: japanese laid-open patent publication No. 2012 and 242377

Disclosure of Invention

In view of the above background, it is desirable to provide a technique for diagnosing whether a coil is good or bad within a wider target range of the coil based on a response voltage obtained by applying a surge voltage.

In view of the above, one embodiment of the present invention provides a diagnostic apparatus for a coil, comprising:

a voltage applying unit that applies an impulse voltage to the coil;

a response voltage detection unit that detects a response voltage from the coil with respect to the surge voltage;

an index calculation unit that calculates a determination index indicating an electrical characteristic of the coil based on the response voltage; and

a determination unit that determines whether or not there is an abnormality in the target coil by comparing the determination index of a reference coil that is a normal coil with the determination index of a target coil that is a diagnostic coil,

as the determination index, in addition to the circuit constant of the coil, at least one of a zero cross point at which the response voltage crosses a predetermined reference voltage and peak voltages on the positive side and the negative side of the response voltage is used.

The technical features of such a diagnostic apparatus for a coil can be applied to a diagnostic method for a coil. For example, the diagnostic method for a coil can have various steps that are characteristic of the diagnostic apparatus having the coil. Of course, the diagnostic method for a coil can achieve the operational effects of the diagnostic device for a coil.

As one embodiment, the method for diagnosing a coil in this case includes the steps of:

a voltage applying step of applying an impulse voltage to the coil;

a response voltage detection step of detecting a response voltage from the coil with respect to the surge voltage;

an index calculation step of calculating a determination index indicating an electrical characteristic of the coil based on the response voltage; and

a determination step of determining whether or not the target coil is abnormal by comparing the determination index of a reference coil which is a normal coil with the determination index of a target coil which is a diagnostic coil,

In the abnormality of the coil to be diagnosed, there is a case where the insulation between conductors constituting the coil is reduced. Even if the insulation between conductors having the same potential is reduced in the coil, the coil has little influence on the electrical performance, and therefore the effect of the circuit constant of the coil as a determination index is low. As is apparent from experiments and simulations performed by the inventors, when the insulation between conductors at the same potential in the coil is reduced, a zero cross point at which the response voltage crosses the reference voltage, and a change in the peak voltage on the positive side or the negative side of the response voltage can be observed. That is, in addition to the circuit constant of the coil, when at least one of a zero cross point at which the response voltage crosses the reference voltage and peak voltages on the positive side and the negative side of the response voltage is used, it is possible to appropriately detect whether the insulation between conductors having the same potential in the coil is reduced or whether the insulation between conductors having different potentials in the coil is reduced. That is, according to the above configuration, the quality of the coil can be diagnosed in a wider target range of the coil based on the response voltage obtained by applying the surge voltage.

Further features and advantages of the diagnostic device for a coil and the diagnostic method for a coil will become apparent from the following description of the embodiments which are described with reference to the accompanying drawings.

Drawings

Fig. 1 is a perspective view showing an example of a stator of a rotating electric machine using coils.

Fig. 2 is a block diagram showing an example of a diagnostic device for a coil in an impact test.

Fig. 3 is an equivalent circuit diagram of the impact test.

Fig. 4 is a waveform diagram showing the response voltage and the sampling principle of the response voltage.

Fig. 5 is an explanatory diagram illustrating the principle of operation of the determination index in response to the voltage/differential voltage/integral voltage.

Fig. 6 is an explanatory diagram showing the principle of operation of the determination index in response to the voltage/differential voltage/second order differential voltage.

Fig. 7 is an explanatory diagram illustrating a principle of discriminating a non-defective product from a defective product.

Fig. 8 is a waveform diagram showing an example of the response voltage.

Fig. 9 is a waveform diagram showing the difference between the response voltage of a non-defective article and the response voltage of a defective article.

Fig. 10 is a winding diagram of a coil assembly of the U-phase.

Fig. 11 is a perspective view of a coil assembly of one phase formed by two concentric windings.

Fig. 12 is a winding diagram of the U-phase and V-phase coil groups.

Fig. 13 is an explanatory diagram illustrating a principle of discriminating non-defective products from defective products in multivariate analysis.

Fig. 14 is an explanatory diagram showing a failure in insulation (short circuit) model and a direction in which a surge voltage is applied.

Fig. 15 is a diagram showing the discrimination between non-defective products and defective products of LC and RC in the case of applying UV.

Fig. 16 is a diagram showing the discrimination between non-defective products and defective products of LC and RC when applied between VWs.

Fig. 17 is a diagram showing the discrimination between non-defective products and defective products of LC and RC in the case of application between WUs.

Fig. 18 is a graph showing discrimination of LC, RC, and zero-crossing time when UV is applied.

Fig. 19 is a diagram showing discrimination of LC, RC, and zero-crossing time when VW is applied.

Fig. 20 is a diagram showing discrimination between LC, RC, and zero-crossing time in the case of WU application.

Fig. 21 is a graph showing the discrimination of LC, RC, and peak voltage when UV is applied.

Fig. 22 is a diagram showing the discrimination of LC, RC, and peak voltage when applied between VWs.

Fig. 23 is a diagram showing discrimination of LC, RC, and peak voltage in the case of application between WUs.

Fig. 24 is a waveform diagram showing the difference between zero-crossing points of non-defective products and defective products.

Fig. 25 is a waveform diagram showing a difference in peak voltage between non-defective products and defective products.

Fig. 26 is a schematic diagram of resistance short circuit estimation.

Fig. 27 is a conceptual diagram of a dead short circuit.

Detailed Description

Hereinafter, embodiments of a diagnostic device for coils of a rotating electric machine or a transformer will be described with reference to the drawings. Here, the diagnostic apparatus 1 for diagnosing whether or not the coil 30 wound around the stator 100 of the rotating electric machine is good as shown in fig. 1 is exemplified. As shown in fig. 1, the stator 100 includes: a core 20 and a coil 30 wound around the core 20. In the present embodiment, the stator 100 generates a rotating magnetic field by three-phase alternating current, and includes three-phase coils corresponding to U-phase, V-phase, and W-phase. The coils 30 of the respective phases are electrically connected at a neutral point N (N1, N2). Core 20 is formed using a magnetic material. A plurality of slots 40 having openings on the inner side in the axial direction AD and the radial direction RD are formed in the core 20 along the circumferential direction CD at a constant rate. The U-phase groove 40, the V-phase groove 40, and the W-phase groove 40 are arranged to repeatedly appear in the circumferential direction CD.

The coil 30 is configured using a linear conductor 35 having electrical conductivity such as copper or aluminum, and an insulating film made of a material having electrical insulation such as resin is formed on the surface of the linear conductor 35. In the present embodiment, the coil 30 formed of the linear conductor 35 of a rectangular flat wire having a cross section in a direction orthogonal to the extending direction is illustrated.

If the insulating coating is insufficient or the insulating coating deteriorates due to damage or the like, the insulation of the linear conductor 35 forming the coil 30 is reduced. As a result, the adjacent linear conductors 35 may be short-circuited to each other, or the linear conductor 35 and the ground may be short-circuited (grounded). The diagnostic apparatus 1 detects insulation failure such as short circuit or grounding as an abnormality of the coil 30, and diagnoses the quality of the coil 30. When the coil 30 is disconnected, the conductors are short-circuited, and the conductors are grounded, the electrical characteristics of the coil 30 greatly vary. However, in the case of a poor insulation, for example, although the resistance value between the adjacent linear conductors 35 varies, the variation in the electrical characteristics of the coil 30 is not significant, and the detection is not easy. Therefore, the presence or absence of insulation failure is determined based on the response voltage generated by applying a surge voltage of a large voltage to the coil 30.

Fig. 2 shows a system configuration of the diagnostic apparatus 1 for diagnosing the coil 30 by the impact test, and fig. 3 shows an equivalent circuit of the impact test. As shown in fig. 2, the diagnostic apparatus 1 includes: a voltage applying unit 2, a response voltage detecting unit (V-DTCT)3, a signal processing unit (SIG-PR)4, a feature value calculating unit (FT-CAL)5, and a determining unit (COMP) 6.

The voltage application unit 2 is a functional unit that applies an impulse voltage to the coil 30, and includes a dc power supply 2a, a current limiting resistor 2b, a capacitor 2c, and a charge/discharge switch 2 d. In a state of being connected to the dc power supply 2a via the charge/discharge switch 2d and the current limiting resistor 2b, electric charges are charged in the capacitor 2 c. Here, when the capacitor 2c and the coil 30 are electrically connected via the charge/discharge switch 2d, the charge charged in the capacitor 2c is continuously discharged to the coil 30 via the charge/discharge switch 2d, and the surge voltage is applied to the coil 30 (fig. 5 and 6: voltage application step # 2).

The response voltage detection unit 3 detects a response voltage from the coil 30 against the surge voltage (fig. 5 and 6: response voltage detection step # 3). As shown in fig. 4, the response voltage detector 3 acquires the response voltage v (t) at time t at predetermined sampling intervals. In the present embodiment, the response voltage detection unit 3 is configured with an a/D converter as a core, and the a/D converter obtains the response voltage by performing analog-to-digital conversion.

As will be described later with reference to FIG. 5, the signal processing unit 4 calculates a differential voltage by differentiating the response voltage and calculates an integral voltage by integrating the response voltage (FIG. 5: signal processing step #4(# 41)). In the present embodiment, the differential voltage and the integral voltage are calculated by differentiating and integrating the response voltage v (t) at each time t acquired by the response voltage detecting unit 3. In the present embodiment, the Signal processing unit 4 is configured with a Processor such as a microcomputer or a DSP (Digital Signal Processor) as a core.

The feature value calculating section 5 calculates a determination index (feature value) indicating the electrical feature of the coil 30 based on the response voltage, the differential voltage, and the integral voltage (fig. 5: feature value calculating step #5(# 51)). The feature amount calculation unit 5 is also configured with a processor such as a microcomputer or a DSP as a core. The feature value calculation step #5 corresponds to an index calculation step. As shown in the equivalent circuit of fig. 3, the coil 30 can be represented as a series circuit of an inductance L and a resistance (resistance component) R. Here, assuming that the line-to-line capacitance of the coil 30 is C, the product LC of the inductance and the capacitance and the product RC of the resistance (resistance component) and the line-to-line capacitance can be used as characteristic quantities representing the electrical characteristics of the coil 30. The feature quantities "LC" and "RC" are determination indexes for determining whether the coil 30 is good or bad by the determination unit 6. Therefore, the feature amount calculation unit 5 can be referred to as an index calculation unit. The characteristic amount (determination index) is not limited to "LC" and "RC", and may be inductance L of coil 30, resistance R of coil, and line-to-line capacitance of coil 30.

The determination unit 6 determines whether or not there is an abnormality in the target coil based on the characteristic amount (determination index) of the target coil, which is the coil 30 to be diagnosed (fig. 5 and 6: determination step # 6). The determination unit 6 is also configured with a processor such as a microcomputer or a DSP as a core. Needless to say, the response voltage detection unit 3, the signal processing unit 4, the feature amount calculation unit 5, and the determination unit 6 may be constituted by, for example, one processor chip incorporating an a/D converter. For example, the determination unit 6 determines the quality of the target coil based on a comparison between the characteristic amount (determination index) of the target coil and the characteristic amount (determination index) of the reference coil, which is the normal coil 30.

As shown in fig. 5, in response voltage detection step #3, response voltage detection unit 3 obtains response voltage v (t) at time t over a period of time t from 0 to n (n: an arbitrary natural number). Of course, the shorter the sampling interval, the higher the time resolution of the response voltage. Further, the resolution of the a/D converter is preferably higher because the voltage resolution of the response voltage is higher. Among them, since the data capacity is increased when the resolution is high, it is preferable to secure a sufficient capacity in a storage element such as a memory or a storage device. Such a memory element or memory device is also included in the response voltage detection unit 3.

The response voltage detection unit 3 sets a range "T" of the response voltage used for the calculation of the feature value. Fig. 8 is a waveform diagram showing an example of a response voltage, and a strain is observed in the response voltage waveform in the initial stage. As described above, the voltage application unit 2 applies the surge voltage by continuously discharging the electric charge stored in the capacitor 2c to the coil 30 via the charge/discharge switch 2 d. At this time, a large current temporarily flows to the charge/discharge switch 2 d. Therefore, in most cases, the charge/discharge switch 2d is formed by connecting a plurality of switching elements in parallel. Since a slight time difference may occur between the switching of the plurality of switching elements, the initial waveform of the response voltage is often disturbed. Therefore, it is preferable that the data for determining the quality of the coil 30 does not include the initial response voltage. In addition, the capacity of the memory element and the memory device is limited. Therefore, as shown in fig. 5, the response voltage detection unit 3 sets a range "T" of the response voltage used for the calculation of the feature quantity to time T (j) to time T (k) (j, k: n or less natural numbers).

As shown in fig. 5, the response voltage in which the valid data range "T" is set in this way can be represented as a matrix having { v (j), v (j +1), v (j +2), ·, v (k) } as an element. In the signal processing step #4(#41), the signal processing section 4 calculates a differential voltage and an integral voltage by differentiating and integrating the response voltage. The differential voltage and the integral voltage can be expressed as a matrix having, as elements, differential values and integral values at respective times t in a range from time t to j to time t to k, as in the response voltage.

The response voltage, the differential voltage, the integral voltage, and the characteristic quantities (LC, RC) can be represented in a matrix form as shown in fig. 5. Here, when a matrix including the response voltage and the differential voltage is "D", a matrix of the feature values (LC, RC) is "X", and a matrix of the integral voltage is "E", the matrix can be represented by the following expression (1).

D·X＝-E···(1)

In order to obtain "X" as a feature amount from expression (1), a transposed matrix D obtained by multiplying "D" on both sides of expression (1) is used^TThen, it becomes formula (2).

D^T·D·X＝-D^T·E···(2)

Then, if the inverse matrix is multiplied on both sides of equation (2) so that only "X" is left on the left side of equation (2), equation (4) is obtained through equation (3). In addition, in the utilization of "D^TWhen the relation between the number of rows and the number of columns of D ″ cannot form an inverse matrix, a pseudo-inverse matrix (pseudo-inverse matrix) is preferably used.

(D^T·D)^-1·D^T·D·X＝-(D^T·D)^-1·D^T·E···(3)

X＝-(D^T·D)^-1·D^T·E···(4)

"X" is a matrix of characteristic quantities (LC, RC) as described above. Therefore, the feature values (LC, RC) can be derived by calculating the right side of the expression (4). That is, the feature values (LC, RC) can be obtained from the acquired response voltages in the procedure shown in the feature value calculation step #5(#51) of fig. 5.

The difference between the non-defective product and the defective product (the coil 30 having the position of the defective insulation), for example, can be distinguished in a two-dimensional space having the elements of the characteristic quantities (LC, RC) as axes by the position and the degree of the defective insulation. That is, the determination unit 6 can determine the non-defective product and the defective product in the two-dimensional space (determination step # 6). Such a determination method is known from japanese patent laid-open No. 2012-242377 mentioned in the description of the background art of the present specification, and therefore, a detailed description thereof is omitted.

In the determination step #6, the determination unit 6 may compare the features of the cluster (group) of the non-defective product (reference coil) with the features of the cluster of the coil to be diagnosed (target coil) (relative recognition). For example, the determination unit 6 may determine a quantitative value such as a distance (distance) or similarity (association). For example, the distance can be an euclidean distance or a mahalanobis distance. In addition to the above, the determination may be performed by using a kernel density function method or a Support Vector Machine (SVM).

However, the derivation of the feature values (LC, RC) is not limited to the form using the response voltage, the differential voltage, and the integral voltage as described above with reference to the signal processing step #4(#41) and the feature value calculation step #5(#51) in fig. 5, and can be derived using the response voltage, the differential voltage, and the second order differential voltage as shown in the signal processing step #4(#42) and the feature value calculation step #5(#52) in fig. 6. The mode shown in fig. 6 is also the same as the mode shown in fig. 5 with respect to the acquisition of the response voltage in the response voltage detection step #3 and the setting of the range "T" of the response voltage used for the calculation of the feature quantity. In the embodiment shown in fig. 5, the signal processing unit 4 calculates a differential voltage and an integral voltage by differentiating and integrating the response voltage (# 41). In the embodiment shown in fig. 6, however, the signal processing unit 4 differentiates the response voltage and calculates a differential voltage and a second order differential voltage by further differentiating the response voltage (# 42). The second order differential voltage can be expressed as a matrix having, as elements, values of second order differentials at respective times t in a range from time t to j to time t to k, as in the response voltage and the differential voltage.

The response voltage, the differential voltage, the second order differential voltage, and the characteristic quantities (LC, RC) can be represented in a matrix form as shown in fig. 6. In the embodiment shown in fig. 5, a matrix including the response voltage and the differential voltage is denoted by "D", a matrix of the feature values (LC, RC) is denoted by "X", and a matrix of the integral voltage is denoted by "E" (# 51). However, in the embodiment shown in fig. 6, "D" is used as a matrix including the differential voltage and the second order differential voltage, and "E" is used as a matrix of the response voltage (# 52). The expressions using "D", "X", and "E" are the same as the above expression (1), and the feature values (LC and RC) can be derived by modifying the expression (1) in the following manner in the same manner as the above expressions (2) to (4).

However, as shown in fig. 6, the waveform of the second order differential voltage is a waveform containing a large amount of noise components because the noise contained in the response voltage is sharpened. Since the second order differential voltage is an element of the matrix "D" and the matrix "D" is also used in expression (4), the noise component affects the accuracy of the characteristic quantities (LC, RC).

Fig. 7 is an explanatory diagram showing a principle of discrimination between non-defective products and defective products, and here, distance (distance) is used as a quantitative value. The left side shows the distribution of GOOD (GOOD) and fail (NG) in the case where the characteristic amount is derived based on the response voltage, the differential voltage, and the second order differential voltage. The right side shows the distribution of GOOD (GOOD) and defective (NG) in the case where the characteristic amount is derived based on the response voltage, the differential voltage, and the integral voltage. These are obtained by schematically depicting the results of simulations based on the inventors. As shown in fig. 7, it is understood that, when the feature value is derived based on the response voltage, the differential voltage, and the integral voltage, the non-defective product (GOOD) and the defective product (NG) are clearly separated from each other, and clear determination can be performed.

However, as described above with reference to fig. 8 and the like, since a strain is observed in the initial stage of the waveform of the response voltage, the response voltage detection unit 3 sets the range "T" of the data so that the initial response voltage is not included in the data for determining the quality of the coil 30. Preferably, the response voltage detection unit 3 sets the range "T" of the data at least after the first zero-crossing point tx1 where the initial response voltage intersects the reference voltage (the center of amplitude, for example, zero volts). That is, preferably, the signal processing unit 4 calculates the differential voltage and the integral voltage based on data after the first zero-crossing point tx1 where the response voltage crosses the reference voltage, and the feature value calculating unit 5 calculates the feature value based on the response voltage after the first zero-crossing point tx1, the differential voltage, and the integral voltage. Of course, the same applies to the case where a second order differential voltage is used instead of the integral voltage.

The amplitude center corresponds to the position of the inflection point in the amplitude center portion of the response voltage waveform. When the response voltage does not include a dc component (offset voltage), the reference voltage corresponds to zero volts. When the response voltage includes a dc component, the reference voltage corresponds to a voltage (offset voltage) of the dc component. Even when the response voltage includes a dc component (offset voltage), zero volts of the ac component can be considered to correspond to the reference voltage. Therefore, regardless of the value of the reference voltage, the point at which the response voltage intersects the reference voltage can be referred to as a "zero-crossing point".

As shown in fig. 8, the response voltage of the vibration is substantially stable after passing through the peak point (first peak point tp1, first positive peak point tpp1 in the present embodiment) of the first time in the positive or negative direction. Therefore, it is preferable to determine the quality of the coil 30 using data after the first zero-crossing point tx1 where the response voltage first crosses the reference voltage after the first peak point tp 1. The peak point indicates the position of a peak, and includes both the position of a peak in the positive direction (a peak in the narrow sense) and the position of a peak in the negative direction (a so-called valley).

In consideration of stability, the response voltage detection unit 3 preferably sets the data range "T" to be after the positive or negative peak point (the second peak point tp2, in the present embodiment, the first negative peak point tnp1) next to the first zero-cross point tx 1. This "T" can be set to, for example, a first period T1 shown in fig. 8. In this case, the signal processing unit 4 calculates the differential voltage and the integral voltage based on data after the second peak point tp2 which is the peak point on the next positive side or negative side of the first zero-crossing point tx1, and the feature quantity calculating unit 5 calculates the feature quantity based on the response voltage after the second peak point tp2, the differential voltage, and the integral voltage. Of course, the same applies to the case where a second order differential voltage is used instead of the integral voltage.

Further, according to the experimental analysis by the inventors, it was confirmed that the difference in the characteristic amount between the non-defective product and the defective product becomes large by further delaying the range "T" of the data. For example, it can be confirmed that the difference in the feature amount becomes larger by changing the range "T" of the data from the first period T1 to the second period T2 in fig. 8. Fig. 9 shows the response voltage of the non-defective product (solid line) and the response voltage of the defective product (broken line) in the first period T1 and the second period T2. Referring to fig. 9, one of the time differences Δ T between the two response voltages in the second period T2 is larger than the time difference Δ T (phase difference) between the two response voltages in the first period T1. Of course, the time difference Δ t between the differential voltage, the integral voltage, or the second order differential voltage is also large, and therefore, the difference between the characteristic values (LC, RC) of the non-defective product and the defective product derived using these is also large.

As is clear from fig. 8 and 9, the amplitude of the response voltage is smaller in the second period T2 than in the first period T1. Therefore, when the voltage resolution is insufficient, the time difference Δ T may increase in the second period T2, while the voltage resolution may decrease and the accuracy may decrease. Therefore, for example, when the resolution of the a/D converter constituting the response voltage detection unit 3 is low, it is preferable that the range "T" of the data to be used is one of the first periods T1 in some cases. In addition, when the reference voltage of the a/D converter is variable and the dynamic range is variable, the dynamic range may be changed in accordance with the attenuation of the amplitude of the response voltage, and the range "T" of the data may be set to the second period T2.

However, as described above, the core 20 has a plurality of slots 40 having openings on the inner side in the axial direction AD and the radial direction RD at constant intervals in the circumferential direction CD. The U-phase groove 40, the V-phase groove 40, and the W-phase groove 40 are arranged to repeatedly appear in the circumferential direction CD. In the present embodiment, the number of the slots 40 for each of the rotor magnetic poles and the stator 100 phases is "2", and the slots 40 for each phase are arranged in the core 20 so as to repeat every two slots in the circumferential direction CD. Fig. 10 shows a winding diagram of a coil assembly of one phase (e.g., U-phase).

As shown in fig. 2 and 10, two concentric windings (the group of U1 and U2 and the group of U3 and U4 in fig. 10) wound around the adjacent slots 40 and connected in parallel are connected in parallel to form one phase. The group of U1 and U2 is, for example, a coil wound in the left-hand direction (CCW) in the circumferential direction CD, and is composed of eight single coils (CCW1, CCW2, CCW3, CCW4, CCW5, CCW6, CCW7, and CCW 8). The group of U3 and U4 is, for example, a coil wound right-handed (CW) in the circumferential direction CD, and is composed of eight single coils (CW1, CW2, CW3, CW4, CW5, CW6, CW7, and CW 8). Fig. 11 illustrates a coil assembly of one phase consisting of two concentric windings.

In such a coil assembly, various insulation failures can occur. For example, in the position "a" shown in the perspective view of fig. 11, there is a possibility that an insulation failure occurs between the linear conductors 35 at the same winding position of the same monocoil, such as the first pattern md1 and the third pattern md3 shown in the winding diagram of fig. 10. For example, in the first mode md1, poor insulation may be generated between the first turn 1T of "U1" and the first turn 1T of "U2" in the first left-handed single coil CCW 1. In addition, in the third mode md3, insulation failure may be generated between the fifth turn 5T of "U1" and the fifth turn 5T of "U2" in the eighth left-handed single coil CCW 8.

In the position "B" shown in the perspective view of fig. 11, there is a possibility that an insulation failure occurs between the linear conductors 35 at different winding positions of the same monocoil, such as the second pattern md2 and the fourth pattern md4 shown in the winding diagram of fig. 10. For example, in the second mode md2, poor insulation may occur between the first turn 1T of "U3" and the second turn 2T of "U4" in the first right-handed single coil CW 1. In addition, in the fourth mode md4, insulation failure may occur between the fifth turn 5T of "U3" and the fourth turn 4T of "U4" in the eighth right-hand single coil CW 8.

Further, as shown in the fifth pattern md5 and the sixth pattern md6 shown in the winding diagram of fig. 10, insulation failure may occur between the linear conductors of the different series of concentric windings. The fifth pattern md5 is an example of insulation failure occurring between the first turn 1T of "U2" in the first left-handed single coil CCW1 and the fifth turn 5T of "U3" in the first right-handed single coil CW 1. In addition, the sixth pattern md6 is an example in which insulation failure occurs between the first turn 1T of "U2" in the first left-handed monocoil CCW1 and the fifth turn 5T of "U3" in the eighth right-handed monocoil CW 8.

In fig. 10 and 11, insulation failure of the coils 30 of the same phase is illustrated, but insulation failure may occur between the linear conductors 35 of different phases. Fig. 12 shows a winding diagram for two phases (here, U-phase and V-phase). As illustrated as a seventh pattern md7 in fig. 12, there is a case where a poor insulation occurs between the fifth turn 5T of "U1" in the eighth left-handed monocoil CCW8 of the U-phase and the fifth turn 5T of "V1" in the eighth left-handed monocoil CCW8 of the V-phase.

The insulation failures in the first to seventh patterns md1 to md7 have been exemplified above. In the first mode md1 and the third mode md3, the positions where the insulation failure occurs are at the same potential, and therefore, as described above, even if it is intended to determine a defective product using LC and RC, which are circuit constants, as the feature quantity, a sufficient difference does not occur. Even if the insulation between the linear conductor 35 and the linear conductor having the same potential is reduced, for example, the electrical performance of the stator 100 is hardly affected. However, from the viewpoint of production control, it is preferable that insulation failure may occur at a position where electrical performance is affected if such an environment with insulation failure occurs in the coil 30. Therefore, it is desirable to be able to appropriately detect such a poor insulation at the same potential position.

According to experiments and simulations of the inventors, it can be confirmed that, as the characteristic amount (determination index), when at least one of a zero cross point at which the response voltage crosses the reference voltage and a peak voltage on the positive side and the negative side of the response voltage is used in addition to the circuit constant of the coil 30 such as the LC and the RC, such an insulation failure at the position can be detected.

As shown in FIG. 8, the zero-crossing points (tx1, tx2, tx3,. tx 13. cndot.) exist at multiple locations. As shown in fig. 9, which is a partially enlarged view, the non-defective product and the defective product are shifted in phase and have different zero-crossing points. Preferably, when the non-defective product is distinguished from the defective product, a characteristic different between the non-defective product and the defective product is extracted. As described above, the initial waveform of the response voltage is often disturbed, and the initial response voltage is preferably not included in the data for determining the quality of the coil 30. Therefore, for example, the zero-cross points used as the feature quantity (determination index) are preferably a plurality of positions after the second zero-cross point tx2 where the response voltage crosses the reference voltage for the second time. Similarly, it is preferable that the peak voltages on the positive side and the negative side of the response voltage used as the feature quantity (determination index) are also voltages after the second cycle of the response voltage (Vp 3 or Vp4 and thereafter shown in fig. 8).

The determination unit 6 performs multivariate analysis of at least three types of feature values (determination indexes) of LC, RC, and zero-crossing points, for example, and determines the quality of the coil 30 based on the distance between the position in the determination coordinate space of the feature value of the reference coil and the position in the determination coordinate space of the feature value of the target coil. Fig. 13 illustrates the results of obtaining distances (distances) as quantitative values by performing multivariate analysis on the first to fifth patterns md1 to md5 and the seventh pattern md 7. The distance can be represented by euclidean distance or mahalanobis distance. As shown in fig. 13, for example, by setting the threshold TH for the distance, it is possible to discriminate non-defective products even in the first pattern md1 and the third pattern md3, which are insulation defects at the same potential. In addition to the above, a kernel density function method or a class SVM can be used for the multivariate analysis of the relative recognition.

In the sixth pattern md6, when the surge voltage is applied, a large current flows because the potential difference is large, and the response voltage cannot be observed, so multivariate analysis is not performed. In the sixth mode md6, a failure of the coil 30 is detected by overcurrent detection.

When the zero-cross point is used as the feature value, the time of each zero-cross point may be used, or a time between two predetermined zero-cross points may be used. In the multivariate analysis illustrated in fig. 13, the time between the second zero-cross point tx2 (reference zero-cross point) and the 14 th zero-cross point tx14 (decision zero-cross point) is used as the feature amount. As described above with reference to fig. 9, since the phase difference between the non-defective product and the defective product tends to increase with the elapse of time, when the time between two zero-crossing points is used as the feature value, it is preferable to set one zero-crossing point at the rear. The front zero-cross point is preferably the second zero-cross point tx2 or later.

The zero-cross points used as the feature quantity (determination index) are a plurality of positions after the second zero-cross point tx2 where the response voltage crosses the reference voltage for the second time. When the time between two zero-cross points is used as the feature value, the zero-cross points at the plurality of positions include a reference zero-cross point set as a zero-cross point after the second zero-cross point tx2 and a determination zero-cross point set as a zero-cross point after the reference zero-cross point. The feature value (determination index) is a time from the reference zero-crossing point to the determination zero-crossing point. In the above example with reference to fig. 8 and 13, the second zero-cross point tx2 is the reference zero-cross point, and the 14 th zero-cross point tx14 is the decision zero-cross point.

In addition, if the time from the reference zero-crossing point to the zero-crossing point is short, the difference between the reference coil and the target coil is unlikely to occur. Therefore, it is preferable that at least one zero-cross point is included as an intermediate zero-cross point between the reference zero-cross point and the decision zero-cross point. In the above example with reference to fig. 8, 11 zero-cross points of the third to 13 th zero-cross points tx3 to tx13 correspond to the intermediate zero-cross points.

As described above, the initial waveform of the response voltage is often disturbed. Therefore, it is preferable that the initial response voltage is not included in the data for determining the quality of the coil 30. The response voltage of the vibration is substantially stable after passing the first peak point in the positive or negative direction (the first peak point tp1, which is the first positive peak point tpp1 in the present embodiment) and the first zero-crossing point tx1 at which the response voltage first crosses the reference voltage. Therefore, it is preferable to determine the quality of the coil 30 using the zero-cross point (zero-cross point after the second zero-cross point tx2) after the peak point (the second peak point tp2, in the present embodiment, the first negative peak point tnp1) in the positive direction or the negative direction next to the first zero-cross point tx 1.

For example, in the mode illustrated in fig. 13, the zero-cross points used as the feature quantities are a plurality of positions (the second zero-cross point tx2 and the 14 th zero-cross point tx14) after the second zero-cross point tx2 at which the response voltage crosses the reference voltage for the second time. According to experiments and simulations by the inventors, the use of one of the second zero-cross point tx2 and the 14 th zero-cross point tx14 improves the determination accuracy, compared to the use of the first zero-cross point tx1 and the 14 th zero-cross point tx 14. Since it is considered that it is preferable to use two points after the second zero-crossing point tx2, it is preferable to set the zero-crossing point according to the examination time, the data capacity that can be accumulated in the diagnostic apparatus 1, and the like.

In the above description, the zero-crossing point is used as the characteristic quantity in addition to the circuit constants (LC, RC) of the coil 30, and for example, as shown in fig. 9, a difference occurs in the amplitude of the voltage between the non-defective product and the defective product. Therefore, in addition to the circuit constants (LC, RC) of the coil 30, the peak voltages on the positive side and the negative side of the response voltage may be used as the characteristic amount. As described above, similarly to the case where the zero-cross point is used as the feature value (determination index), the peak voltages on the positive side and the negative side of the response voltage are also preferably the voltages at the peak points after the second cycle of the response voltage (the third peak voltage vp3 (the second positive peak vpp2) at the third peak point tp3 (the second positive peak point tpp2) or the peak voltages appearing after the fourth peak voltage vp4 (the second negative peak voltage vnp2) at the fourth peak point tp4 (the second negative peak point tnp2) shown in fig. 8). The peak voltage used as the feature value is preferably one or more of the fifth negative peak point tnp5 to the eighth positive peak point tpp8 (tnp5, tpp6, tnp6, tpp7, tnp7, tpp8) included in the second period T2 in fig. 9, for example.

The "voltage at the peak point after the second period" is "a voltage at the peak point after at least two peak points have passed". The "second period" is preceded by a "first period" in which at least two peak points are present. The third peak point is a peak point of the "second period", so that the "voltage of the peak point after the second period" is a "voltage of the peak point after passing at least two peak points". In the example shown in fig. 8, the first peak point tp1 (first positive peak point tpp1) and the second peak point tp2 (first negative peak point tnp1) are peak points of a "first cycle". Further, the third peak point tp3 (second positive peak point tpp2) and the fourth peak point tp4 (second negative peak point tnp2) are peak points of the "second cycle". Therefore, the peak points after the second period are the peak points after the third peak point tp3 (second positive peak point tpp 2).

Alternatively, in addition to the circuit constants (LC, RC) of the coil 30, the four characteristic quantities may be subjected to multivariate analysis using the zero-crossing point and the peak voltage as the characteristic quantities. In the case of performing multivariate analysis using four feature quantities, in addition to the circuit constants (LC, RC) of the coil 30, four feature quantities, i.e., the time between the second zero-cross point tx2 and the 14 th zero-cross point tx14 and the time between the first zero-cross point tx1 and the 14 th zero-cross point tx14, may be used. In addition to the circuit constants (LC, RC) of the coil 30, five characteristic quantities of the peak voltage, the time between the second zero-cross point tx2 and the 14 th zero-cross point tx14, and the time between the first zero-cross point tx1 and the 14 th zero-cross point tx14 may be used.

In deriving the circuit constants (LC, RC) of the coil 30, it is preferable to calculate based on the response voltage, the differential voltage, and the integral voltage as described above, but calculation based on the response voltage, the differential voltage, and the second order differential voltage is also possible.

Hereinafter, the usefulness of using the zero-cross point and the peak voltages on the positive side and the negative side of the response voltage as the feature value (determination index) is described above, but experiments and simulations performed by the inventors are described below.

Fig. 14 shows the experimental results illustrated below, a poor insulation model (short circuit model) corresponding to the simulation results, and the application direction of the surge voltage. Here, the insulation failure (short circuit) in the seventh pattern md7 as illustrated with reference to fig. 12 was modeled. That is, the case where the eighth left-hand single coil CCW8 of the U-phase (U1) and the eighth left-hand single coil CCW8 of the V-phase (V1) are short-circuited is modeled. The symbol "Rs" shows the short-circuit resistance at the time of this short circuit. Here, a case where the short-circuit resistance Rs is substantially zero is referred to as a "dead short". In addition, the substantially zero indicates a case where the resistance value of the short-circuit resistance Rs is smaller than the short-circuit threshold value, and the short-circuit threshold value is preferably smaller than the impedance (resistance value) of the single coil (for example, smaller than 10[ m Ω ]). When the short-circuit resistance Rs is equal to or higher than the short-circuit threshold, for example, a case of 10 m Ω or more is referred to as "resistance short circuit" here. Hereinafter, the results of experiments and simulations performed with "Rs ═ 0" in the case of a dead short circuit and with various values "Rs > 0" in the case of a resistance short circuit will be described.

As for the surge test using the diagnostic apparatus 1, as described above with reference to fig. 2, when the coil 30 is diagnosed, a surge voltage is applied to each of the U-V inter-phase (IM1), the V-W inter-phase (IM2), and the W-U inter-phase (IM3), as shown in fig. 14. Fig. 15 to 16 show distances (distances) between non-defective products and defective products when LC and RC are characteristic quantities when an impact voltage is applied. The distance is the euclidean distance or the mahalanobis distance as described above. Fig. 15 shows the distance when the U-V phases are applied, fig. 16 shows the distance when the V-W phases are applied, and fig. 17 shows the distance when the W-U phases are applied. Here, 14 kinds of resistance values are used as the short-circuit resistance Rs. The resistance value on the left side of the horizontal axis is small (e.g., 100[ m Ω ]), and the resistance value on the right side is large (e.g., 10[ Ω ]).

As is clear from fig. 15 to 17, in the case of a complete short circuit, the distance is different from the non-defective product, and the non-defective product can be identified by setting an appropriate threshold value. However, in the case of a short-circuit in the resistance, there is no significant difference in distance between the non-defective product and the defective product, and it is difficult to distinguish the defective product even if a surge voltage is applied to any of the phases. As described above, there is a limit to diagnosis in the case where LC and RC are used as feature amounts (determination indexes).

Fig. 18 to 20 show the distances between non-defective products and defective products when LC, RC, and zero-crossing time are used as feature quantities. Fig. 18 shows the distance when the U-V interphase is applied, fig. 19 shows the distance when the V-W interphase is applied, and fig. 20 shows the distance when the W-U interphase is applied. The horizontal axis of fig. 18 to 20 is the same as fig. 15 to 17. In this case, in the case of a dead short, the distance is different from the non-defective product, and the non-defective product can be identified by setting an appropriate threshold value. However, in the case of a short-circuit in the resistance, there is no significant difference in distance between the non-defective product and the defective product, and it is difficult to distinguish the defective product even if a surge voltage is applied to any of the phases.

Fig. 21 to 23 show the distances between non-defective products and defective products when LC, RC, and peak voltage are used as the feature values. The peak voltage of the eighth period denoted by "E8" in fig. 8 is used here. Fig. 21 shows the distance when the U-V interphase is applied, fig. 22 shows the distance when the V-W interphase is applied, and fig. 23 shows the distance when the W-U interphase is applied. The horizontal axis of fig. 21 to 23 is the same as that of fig. 15 to 21. As shown in fig. 21, when a surge voltage is applied to the U-V phase, and LC, RC, and the peak voltage are used as characteristic quantities, the distance is different from that of a non-defective product in the case of a dead short circuit or in the case of a resistance short circuit. Therefore, by setting an appropriate threshold value, defective products can be identified.

When the surge voltage is applied to the V-W phase, as shown in fig. 22, the difference in distance between the non-defective product and the defective product is smaller than that when the surge voltage is applied to the U-V phase. However, since the distance is different from that of the non-defective product in the case of a dead short or in the case of a resistance short, the non-defective product can be identified by setting an appropriate threshold value.

On the other hand, when the surge voltage is applied to the W-U phase, the distance is different from that of the non-defective product in the case of the complete short circuit, but the distance between the non-defective product and the defective product is not significantly different in the case of the resistance short circuit. When a surge voltage is applied to the W-U phase, a defective product can be identified with respect to a dead short, but a defective product is difficult to identify with respect to a resistance short. In the diagnosis, since the surge voltage is applied to each of the U-V phase (IM1), the V-W phase (IM2), and the W-U phase (IM3) in this order, defective products can be identified by any of them.

As described above with reference to fig. 18 to 20, when the abnormality of the target coil is a dead short, the determination unit 6 can appropriately determine whether or not the abnormality of the target coil exists, using the zero-cross point as the determination index. As described above with reference to fig. 21 to 23, when the abnormality of the target coil is a resistance short circuit, the determination unit 6 can appropriately determine whether or not the abnormality of the target coil exists, using the peak voltage as the determination index.

Fig. 24 is a waveform diagram showing the difference in zero-crossing points of the non-defective products and the defective products in the eighth period (E8) of the response voltage shown in fig. 8. The solid line shows the response voltage of the non-defective product, and the other two show the response voltages of the subject coil in the dead short state. As shown in fig. 24, the zero-crossing point "tp 8 s" of the response voltage of the target coil in the dead short state is earlier than the zero-crossing point "tp 8 g" of the response voltage of the non-defective product. Therefore, when the abnormality of the target coil is a dead short, the determination unit 6 can appropriately determine whether or not the abnormality of the target coil exists, by using the zero-cross point as the determination index.

Fig. 25 is a waveform diagram showing the difference in peak voltage between non-defective products and defective products in the eighth period (E8) of the response voltage shown in fig. 8. The solid line shows the response voltage of the non-defective product, and the other two lines show the response voltages of the subject coils in the resistance short-circuit state. As shown in fig. 25, the peak voltage "Vpp 8 s" of the response voltage of the target coil in the resistance short-circuited state is lower than the peak voltage "Vpp 8 s" of the response voltage of the non-defective product. Therefore, when the abnormality of the target coil is a resistance short circuit, the determination unit 6 can appropriately determine whether or not the abnormality of the target coil exists, by using the peak voltage as the determination index.

Fig. 26 shows a conceptual diagram of a resistance short circuit. Specifically, a case where a resistance short circuit occurs in the eighth left-hand single coil CCW8 of the U-phase single system (U1) is exemplified. The resistance component of the coil 30 is shown here as an equivalent circuit. In this case, the short-circuit current Is In the direction opposite to the phase current In flowing to the U-phase (U1) flows from another position connected by the resistance short circuit, and interferes with the phase current In. As a result, the speed of the current (time derivative of the current) "di (t)/dt" of the coil 30 in which the resistance short circuit occurs is smaller than that of the acceptable product. The voltage drop "e (t)" in the coil 30 having the inductance L is represented by the product of the speed of the current and the inductance L. That is, "e (t) ═ L (di (t)/dt)". Since the speed "di (t)/dt" of the current of the coil 30 in which the resistance short occurs is smaller than that of the non-defective product, the voltage drop "e (t)" of the coil 30 in which the resistance short occurs is also smaller than that of the non-defective product.

Fig. 27 shows a speculative schematic of a dead short. Specifically, the example shows a case where a dead short circuit occurs in the eighth left-hand single-coil CCW8 of one system of U-phase (U1). As described above, the impedance at the position where the dead short occurs (the resistance value of the short-circuit resistance Rs) is smaller than the impedance of the monocoil, so the phase current In hardly flows to the eighth left-handed monocoil CCW8 having a larger impedance than the short-circuited position. As a result, the number of single coils of the U-phase system (U1) is reduced, and the inductance L of the coil 30 in which the dead short occurs is smaller than that of the acceptable product. As a result, the resonance frequency "fc" represented by the following formula (5) becomes high.

fc＝1/(2π·((LC)^(1/2)))···(5)

The resonance frequency fc becomes high, and thus the period of the response voltage becomes short, so the zero-crossing point occurs in advance.

[ brief description of the embodiments ]

The outline of the diagnostic apparatus (1) for a coil described above will be briefly described below.

A diagnostic device (1) for a coil, as one embodiment, is provided with:

a voltage application unit (2) that applies a surge voltage to the coil (30);

a response voltage detection unit (3) that detects a response voltage from the coil (30) with respect to the surge voltage;

an index calculation unit (5) that calculates a determination index indicating an electrical characteristic of the coil based on the response voltage; and

a determination unit (6) that determines whether or not there is an abnormality in the target coil by comparing the determination index of a reference coil that is a normal coil (30) with the determination index of the target coil that is a diagnostic coil (30),

the determination index is a value obtained by using, in addition to the circuit constant of the coil (30), at least one of a zero cross point at which the response voltage crosses a predetermined reference voltage and peak voltages on the positive side and the negative side of the response voltage.

The technical features of the diagnostic device (1) for a coil can also be applied to a diagnostic method for a coil. For example, the diagnostic method for a coil can have various steps of the characteristics of the diagnostic device (1) having the coil. Of course, the diagnostic method of the coil can also realize the operation and effect of the diagnostic device (1) of the coil.

As one embodiment, the diagnostic method for a coil in this case includes the steps of:

a voltage application step (#2) of applying a surge voltage to the coil (30);

a response voltage detection step (#3) for detecting a response voltage from the coil (30) to the surge voltage;

an index calculation step (#5) of calculating a determination index indicating an electrical characteristic of the coil based on the response voltage; and

a determination step (#6) of determining whether or not there is an abnormality in the target coil by comparing the determination index of a reference coil of the coil (30) that is normal with the determination index of the target coil of the coil (30) that is a diagnosis target,

In the abnormality of the coil to be diagnosed, there is a case where the insulation between conductors constituting the coil is reduced. In the coil (30), even if the insulation between conductors having the same potential is reduced, the coil has little influence on the electrical performance, so that the effect of the circuit constant of the coil (30) as a determination index is reduced. Experiments and simulations carried out by the inventors have revealed that, when the insulation between conductors at the same potential in the coil (30) is reduced, a zero-crossing point at which the response voltage crosses the reference voltage, and a change in the peak voltage on the positive side or the negative side of the response voltage can be observed. That is, in addition to the circuit constant of the coil (30), when at least one of a zero cross point at which the response voltage crosses the reference voltage and peak voltages on the positive side and the negative side of the response voltage is used, it is possible to appropriately detect whether the insulation between conductors having the same potential in the coil (30) is reduced or whether the insulation between conductors having different potentials in the coil (30) is reduced. That is, according to the above configuration, the quality of the coil (30) can be diagnosed in a wider target range of the coil (30) based on the response voltage obtained by applying the surge voltage.

Here, the circuit constant of the coil (30) preferably includes: a product (LC) of an inductance (L) of the coil (30) and a line-to-line capacitance (C) of the coil (30), and a product (RC) of a resistance (R) of the coil (30) and the line-to-line capacitance (C).

The circuit constant of the coil (30) is, for example, the inductance (L) of the coil (30), the resistance (R) of the coil (30), the line-to-line capacitance (L) of the coil (30), and the product (LC, RC, LR) thereof. When independent circuit constants are derived from the response of the coil (30) to electrical stimulation in the application of surge voltage or the like, the calculation is often complicated. On the other hand, the product of the independent circuit constants corresponds to the combined impedance of the coil (30), so that the product can be easily derived from the response to the electrical stimulation as compared with the case where the independent circuit constants are derived. Further, the accuracy of the derived synthetic impedance (product of circuit constants) tends to be high also from the relationship between the synthetic impedance and the response voltage. Therefore, it is preferable that the determination index includes a product (LC) of the inductance (L) and the line-to-line capacitance (C) and a product (RC) of the resistance (R) and the line-to-line capacitance (C).

Preferably, the reference voltage is zero volts.

If the reference voltage is zero volts, the calculation becomes easy. The reference voltage serving as a reference for the zero-cross point is substantially the center of the amplitude of the response voltage and corresponds to the position of the inflection point in the center of the amplitude of the response voltage. When the response voltage does not include a dc component (offset voltage), the reference voltage corresponds to zero volts. Such a dc component can be removed by using a coupling capacitor or the like, for example, and therefore, the reference voltage can be set to zero volts with a simple configuration.

Here, the zero-cross points are preferably a plurality of positions after a second zero-cross point (tx2) at which the response voltage crosses the reference voltage for the second time.

For example, the voltage application unit (2) applies a surge voltage by continuously discharging the charge stored in the capacitor (2c) to the coil (30) via the switch (2 d). At this time, since a large current flows to the switch (2d), the switch (2d) is often configured by connecting a plurality of switching elements in parallel. Since there is a slight time difference in switching of the plurality of switching elements, the initial waveform of the response voltage is often disturbed. In addition, when the switch (2d) is formed of a single switching element, the initial waveform of the response voltage may be disturbed by the occurrence of a buzzer (ringing) or the like. Therefore, it is preferable that the data for determining the quality of the coil (30) does not include the initial response voltage. The response voltage of the vibration is substantially stabilized after passing the peak point (tp1) of the first time in the positive or negative direction and passing the first zero-crossing point (tx1) where the response voltage first crosses the reference voltage. Therefore, it is preferable to determine the quality of the coil (30) by using the zero-cross point (zero-cross point after the second zero-cross point (tx 2)) after the peak point (tp2) in the positive direction or negative direction next to the first zero-cross point (tx 1).

Here, the zero-cross points at the plurality of positions preferably include: a reference zero-cross point set to the zero-cross point after the second zero-cross point (tx2), and a determination zero-cross point set to the zero-cross point after the reference zero-cross point, wherein the determination index is a time from the reference zero-cross point to the determination zero-cross point.

Experiments and simulations performed by the inventors have revealed that a phase difference occurs in the response voltage between a non-defective product and a defective product. The phase difference also affects the period of the response voltage, and the period (including a half period and a total value of a plurality of periods) can be obtained by the time between different zero-crossing points. Therefore, according to the present configuration, the quality of the coil (30) can be appropriately diagnosed.

Preferably, at least one zero-cross point is included as an intermediate zero-cross point between the reference zero-cross point and the determination zero-cross point.

As is clear from experiments and simulations performed by the inventors, the phase difference between a non-defective product and a defective product tends to increase with the passage of time. When the time between two zero-cross points is used as the determination index, one zero-cross point is preferably set to the rear. That is, it is preferable to set the zero-cross point including another zero-cross point (intermediate zero-cross point) rather than the case where the reference zero-cross point and the determination zero-cross point are adjacent zero-cross points.

Here, the peak voltage is preferably a voltage after the second period of the response voltage.

The voltage of the first period of the response voltage may be distorted in the voltage waveform and the voltage value may become unstable due to the configuration of the diagnostic device (1) such as the voltage application unit (2) for applying the surge voltage and the response voltage detection unit (3) for detecting the response voltage. For example, the voltage application unit (2) is often configured by connecting a plurality of switching elements in parallel. When a slight time difference occurs in switching of the plurality of switching elements, the initial waveform of the response voltage is disturbed. Therefore, the initial response voltage is not used to determine the quality of the coil (30), and the peak voltage after the second period with a more stable voltage is used, thereby improving the reliability of diagnosis.

Here, the abnormality of the target coil preferably includes an insulation failure between conductors (35) constituting the target coil.

The abnormality of the target coil includes disconnection, short circuit with the ground, poor insulation between conductors (35) constituting the coil (30), and the like. Disconnection, short circuit with the ground, and the like can also be detected relatively clearly by other test methods. In the poor insulation between the conductors (35), although the resistance value between the conductors varies, the detection is difficult to distinguish from the measurement error in the measurement of the resistance value, for example. However, the variation in the resistance value due to the insulation failure can be detected also from the response voltage with respect to the applied surge voltage. Therefore, the abnormality of the target coil diagnosed by the coil diagnosis device (1) preferably includes insulation failure between conductors (35) constituting the coil (30).

Here, it is preferable that the determination unit determines whether or not there is an abnormality in the target coil by using the zero-cross point as the determination index when the abnormality in the target coil is a dead short circuit in which the target coil is short-circuited at a resistance value smaller than a predetermined short-circuit threshold value, and by using the peak voltage as the determination index when the abnormality in the target coil is a resistance short circuit in which the target coil is short-circuited at a resistance value equal to or larger than the short-circuit threshold value.

As is clear from experiments and simulations performed by the inventors, in the case of a dead short circuit, when the zero-crossing point is used as a determination index, the difference between the non-defective product and the coil having an abnormality becomes clear. In the case of a resistance short circuit, it is found that the difference between the non-defective coil and the coil having an abnormality is clear when the peak voltage is used as the determination index. That is, by appropriately using the determination index, it is possible to diagnose the abnormality of the target coil including the type of short circuit.

Preferably, the determination unit (6) performs at least three types of multivariate analysis of the determination index, and performs the determination based on a distance between a position in a determination coordinate space of the determination index of the reference coil and a position in the determination coordinate space of the determination index of the target coil.

In multivariate analysis, events represented by a plurality of parameters can be quantitatively represented, and it is suitable for separation of events that can be distinguished by a plurality of parameters. Therefore, the determination unit (6) performs multivariate analysis, thereby diagnosing whether the coil is good or bad in a wider target range of the coil.

Preferably, the diagnostic device (1) for a coil further includes a signal processing unit that calculates a differential voltage by differentiating the response voltage and calculates an integral voltage by integrating the response voltage, and the index calculation unit (5) calculates the determination index indicating the electrical characteristic of the coil (30) based on the response voltage, the differential voltage, and the integral voltage.

There is known a method of calculating a determination index indicating an electrical characteristic of a coil (30) based on a response voltage, a differential voltage, and a second order differential voltage, and determining whether the coil is good or bad. However, since the order of differentiation is increased, the frequency of high noise components is also sharper, and therefore, the accuracy of the determination index is lowered, and the accuracy of the quality determination of the coil (30) may also be lowered. In the present configuration, the determination index is calculated based on the response voltage, the differential voltage, and the integral voltage. The number of divisions of the order of three voltages is "2" in both the case of using the response voltage, the differential voltage, and the second order differential voltage, and the case of using the response voltage, the differential voltage, and the integral voltage. That is, in the case of using the response voltage, the differential voltage, and the integral voltage, the differential voltage is a second derivative as viewed from the integral voltage. Therefore, the determination index can be appropriately calculated based on the response voltage, the differential voltage, and the integral voltage. In this case, since the sharpening of the noise component is suppressed with an increase in the order of the differentiation, a decrease in the accuracy of the determination index can be suppressed. As a result, the accuracy of the determination of the quality of the target coil based on the determination index can be improved. That is, according to the present configuration, the quality of the coil (30) can be diagnosed with higher determination accuracy based on the response voltage obtained by applying the surge voltage.

Description of reference numerals

1: a diagnostic device (diagnostic device for coil); 2: a voltage applying section; 3: a response voltage detection section; 4: a signal processing unit; 5: a feature value calculation unit (index calculation unit); 6: a determination unit; 30: a coil; 35: a linear conductor (conductor); tx 2: a second zero crossing point; v (t): a response voltage; LC: a feature value (determination index); RC: a feature value (determination index); x: a feature value (determination index); # 2: a voltage application step; # 3: a response voltage detection step; # 5: a feature value calculation step (index calculation step); # 6: and a judging step.

Claims

1. A diagnostic device for a coil, comprising:

a voltage applying unit that applies an impulse voltage to the coil;

2. The diagnostic apparatus of a coil according to claim 1,

the circuit constant of the coil includes: a product of an inductance of the coil and a line-to-line capacitance of the coil, and a product of a resistance of the coil and the line-to-line capacitance.

3. The diagnostic apparatus of coil according to claim 1 or 2,

the reference voltage is zero volts.

4. The diagnostic device for coil according to any one of claims 1 to 3,

the zero-cross points are a plurality of positions after a second zero-cross point at which the response voltage crosses the reference voltage for the second time.

5. The diagnostic apparatus of a coil according to claim 4,

the zero-crossing points of the plurality of positions include: a reference zero-cross point set to the zero-cross point after the second zero-cross point, and a determination zero-cross point set to the zero-cross point after the reference zero-cross point,

the determination index is a time from the reference zero-crossing point to the determination zero-crossing point.

6. The diagnostic apparatus of a coil according to claim 5,

at least one zero-crossing point is included as an intermediate zero-crossing point between the reference zero-crossing point and the decision zero-crossing point.

7. The diagnostic device for coil according to any one of claims 1 to 6,

the peak voltage is a voltage after the second period of the response voltage.

8. The diagnostic device for coil according to any one of claims 1 to 7,

the abnormality of the target coil includes an insulation failure between conductors constituting the target coil.

9. The diagnostic apparatus of a coil according to claim 8,

when the abnormality of the target coil is a dead short in which the target coil is short-circuited by a resistance value smaller than a predetermined short-circuit threshold value, the zero-cross point is used as the determination index,

when the abnormality of the target coil is a resistance short-circuit in which the target coil is short-circuited by a resistance value equal to or greater than the short-circuit threshold, the determination unit determines whether or not the target coil is abnormal, using the peak voltage as the determination index.

10. The diagnostic device for coil according to any one of claims 1 to 9,

the determination unit performs at least three types of multivariate analysis of the determination index, and performs the determination based on a distance between a position in a determination coordinate space of the determination index of the reference coil and a position in the determination coordinate space of the determination index of the target coil.

11. The diagnostic device for a coil according to any one of claims 1 to 10, further comprising:

a signal processing unit for differentiating the response voltage to calculate a differential voltage and integrating the response voltage to calculate an integral voltage,

the index calculation unit calculates the determination index indicating the electrical characteristic of the coil based on the response voltage, the differential voltage, and the integral voltage.

12. A diagnostic method for a coil, comprising the steps of:

a voltage applying step of applying an impulse voltage to the coil;